[From Bruce Abbott (971218.1145 EST)]
Bill Powers (971218.1913 MST) --
Can we get back to talking about PCT now?
Sounds good. For starters, let's review all the important discoveries about
mind and behavior that have been made through documented empirical PCT
research over the past 50 years.
Regards,
Bruce
i.kurtzer (971219.0000)
[From Bruce Abbott (971218.1145 EST)]
Bill Powers (971218.1913 MST) --
Can we get back to talking about PCT now?
Sounds good. For starters, let's review all the important discoveries about
mind and behavior that have been made through documented empirical PCT
research over the past 50 years.
1. People can and do control.
2. In a control task there is an insignificant correlation between the output
and input.
3. For those tasks the best predictor of the controlled variable was the one
that correlated the least with the disturbance.
4. Control is articulated about the input stream.
5. Control can occur in a variety of nasty environs.
6. For those tasks that are control tasks only one theory has been predictively
meaningful.
7. Parameter adjustment of the working model to a particular person are
predictively successful over 5 years.
8. Since science is suppose to progress through theories that are a predictive
improvement, then science does not progress that way.
as an overview, for the nitty-gritty read anything by Bourbon, Marken, and
Powers.
i.
[From Tim Carey (971219.1650)]
[From Bruce Abbott (971218.1145 EST)]
Sounds good. For starters, let's review all the important discoveries
about
mind and behavior that have been made through documented empirical PCT
research over the past 50 years.
We could ... but wouldn't that be a bit droll. For me, the discovery that
we control perceptual input not motor output says it all. Let's talk about
some other stuff, like applying the MOL to different situations.
Regards,
Tim
[From Bill Powers (971219.0535 MST)]
Bruce Abbott (971218.1145 EST)--
Sounds good. For starters, let's review all the important discoveries about
mind and behavior that have been made through documented empirical PCT
research over the past 50 years.
That would take us back to about 1948, which isn't a bad span to consider,
although it's five years before I got interested in control theory and
started to learn it. Sixty or 65 years would take us back to the 1930s,
when electrical engineers were just starting to work out the principles of
control with electromechanical systems. That might be an even better place
to start.
The main documented discoveries during these earliest years concerned
things that we now rather take for granted; the basic functions required
for control, such as sensing, comparison with reference signals, and
conversion of error signals to output with suitable dynamic filtering to
achieve stability. In these early years, all the design work was aimed at
automating tasks formerly done by human beings, such as adjusting
voltage-balancing circuits to read the voltage of a standard cell, or
adjusting generator outputs of power plants to compensate for changes in
the load, or aiming guns on board a rolling ship. All these tasks were of
one particular kind: they involved variable actions that had to be adjusted
on the fly to maintain something in the environment in a constant condition
or bring something to some desired state and keep it there, in the presence
of unpredictable disturbances. Open-loop machines of the kind that had been
prevalent since before the Industrial Revolution could not do this sort of
thing; only human beings could. This is why all those old machines had
human operators: not to act like another cog in the industrial machine, but
to do all those variable and unpredictable things that needed to be done to
keep the classical machine working right. Charlie Chaplin's "Modern Times"
captured the picture of the human cog in the machine, but missed the fact
that only a human being could have put a wrench on a passing bolt and
tighten it by just the amount needed.
In developing these surrogates for the human operator, the early electrical
engineers were actually building models of human behavior, although they
didn't consider that their primary objective. So I think we could say that
the real research on PCT began with these engineers, who of course left
behind a great deal of documentation.
During World War II, cybernetics began to form and culminated in Wiener's
1948 book. At the same time, many "engineering psychologists" began to
investigate the human operator, for the first time representing what the
human being in a man-machine system was doing explicitly as a control
process. Biologists and physiologists also started using analog computing
as a way of simulating control systems within the body: biochemical
systems, reflexes, and other regulatory processes. The first steps were
taken toward analyzing neuromuscular control systems. There was a fairly
extensive literature in all these fields, although the technical nature of
the analyses restricted the number of people who could understand what they
meant. As an aside, Bandura once published a diagram of an analog computing
setup being used by, I think, some British investigators to analyse a
physiological system. Bandura used it as an example of a model that nobody
could possibly understand.
I was inspired by reading Wiener's book in about 1952. Wiener had mentioned
the fact the some control loops might extend outside the body, so that
control theory could apply to more than balancing the body or maintaining
body temperature. By 1953 I had realized that feedback control could be
involved in all aspects of behavior, and I began to learn about it, aided
by analog computers which themselves contained many lessons about how
feedback works. I learned about control of input not from servomechanism
theory but from reading about how the operational amplifiers in an analogue
computer worked. The critical comment made by one writer was that an
operational amplifier can most easily be understood as maintaining its
negative input at the same voltage as its positive input, via the feedback
path where the computing components were placed. When I read that I
suddenly understood how human control systems work, and that was the true
beginning of PCT.
During the 1950s, with Bob Clark, I did my first experiments with human
control systems. The equipment was crude and the ideas often misguided, but
gradually a picture started to form of a hierarchy of control, and I began
to see how it might be possible to simulate control and thus evaluate the
parameters of real human control systems. I spent a lot of time building
gadgets to measure control properties. One of them still sits in my storage
shed -- the original of the apparatus Dick Robertson later emulated with a
computer and used to do the work on nested reaction times in his "Phantom
Plateau" paper. In those days only oscilloscopes and strip-chart or X-Y
recorders were available for displaying controlled variables and recording
the results of experiments; my analog computer got hooked into many of
these experiments as a handy way of providing the control electronics. And
I built many a rat's nest of vacuum-tube circuits and many a
cobbled-together pneumatic actuator, the equipment and design time being
unintentionally subsidized by the V. A. Research Hospital where all this
took place. It was here, by the way, that I established the true reaction
time in a spinal motor control system as being something under 50
milliseconds; this was done by hooking up an electromyograph to pronator
teres and finding the earliest correlated spike after a mechanical
disturbance of a handle that a person was holding steady in torsion gainst
mechanical disturbances (that measure, it turns out, is much too long).
Even then, psychologists were saying that feedback is too slow, and I
wanted to show that it isn't.
There were many other experiments along the way. If any of them had been
picked up by others, the way Skinner's little "cumulative record" device
got picked up and used, there would doubtless be a large literature
relating to them. But of course in the late 1950s, when we were trying to
get psychologists interested in these things, the digital computer was
already on the scene and starting to take over, and there was little
interest in our arcane explorations. It wsn't long before everyone more or
less forgot how to "do analog."
After the 1960 papers, Clark, MacFarland, and I split up and I went off to
try to get a degree in psychology because Don Campbell pulled some strings
and got me a scholarship at Northwestern. I lasted one year in that
Spencian department, and then went to work at the Dearborn Observatory
building low-light-level TV cameras for astronomy. I was pretty well burned
out on control theory, especially with respect to getting any psychologists
to take it seriously. I gave one paper at a biophysical society meeting,
then ten years later published the paper (the Verhave rat experiment) in
_Behavioral Science_, but that was all I had the heart to do. If it weren't
for Don Campbell (again!) I suppose I might have eventually given up on the
whole thing. But he kept after me to write the book that became B:CP, and
he and Hugh Petrie arranged for me to give talks at lunchtime seminars and
give a student-sponsored seminar, and just generally wouldn't let me quit.
So in 1973 B:CP came out and I looked forward to becoming rich and famous.
After the book, I gave lots of seminars and wrote lots of papers for
various journals, most by invitation, and started meeting the people who
are still interested in PCT -- and scads more who weren't. I acquired a
PDP8-e and started trying to model tracking behavior again, the model
getting simpler and simpler and fitting the data better and better. I wrote
the Byte articles, and the Spadework article where I published some of the
results of the computer experiments (I had to trace the computer displays
using a half-silvered mirror, since plotters were beyond my means). Still
no spark.
Well, the rest is recent history. The documented discoveries are mostly
pretty simple, showing how people can control different kinds of visual
variables and sound variables. Marken, Bourbon, Robertson, and Goldstein,
and of course Tom Bourbon's students and others, have all done and
published PCT experimental studies. There are some people, notably Tim
Carey, putting the ideas about levels of control into practice, but
documentation is still to come.
I think that the kernel of documented discoveries is there. Any one of them
could be used as a jumping-off point for all kinds of research, should
there be a surge of interest in PCT. Skinner's revolution consisted
primarily of hundreds of people doing the same experiment over and over in
dozens of different labs, each experimenter adding his own twists and
exploring his own ideas. The same thing would happen if a lot of people
caught on to PCT and starting ringing the changes, looking for new
phenomena. The fact that this hasn't happened in PCT is an accident of the
times, a result of competition with other ideas that became established
much earlier and were more compatible with traditional thinking.
Best,
Bill P.
[From Rick Marken (971219.0830)]
Bruce Abbott (971218.1145 EST)
For starters, let's review all the important discoveries about
mind and behavior that have been made through documented empirical
PCT research over the past 50 years.
I think this is a good question and I think you've already gotten
some good answers. Before I give you my answer I would like to
hear your answer to your own question; what do you think are the
important discoveries about mind and behavior that have been made
through documented empirical PCT research over the past 50 years?
Since yours is likely to be a very short answer to this question,
perhaps you might also add some examples of the important discoveries
about mind and behavior that have been made through documented
empirical _non-PCT_ research over the past 50 years. I don't think
this would be inconsistent with Bill's "That's nice. Now can we talk
about PCT?" policy since important discoveries about mind and
behavior are certainly releveant to PCT.
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Tim Carey (971220.0600)]
[From Bill Powers (971219.0535 MST)]
That would take us back to about 1948, which isn't a bad span to
consider,
although it's five years before I got interested in control theory and
started to learn it. Sixty or 65 years would take us back to the 1930s,
when electrical engineers were just starting to work out the principles
of
control with electromechanical systems. That might be an even better
place
to start.
A great post Bill, this will be a really useful reference. Thanks
Tim
[From Bruce Abbott (971223.1020 EST)]
Rick Marken (971219.0830) --
Bruce Abbott (971218.1145 EST)
For starters, let's review all the important discoveries about
mind and behavior that have been made through documented empirical
PCT research over the past 50 years.
I think this is a good question and I think you've already gotten
some good answers. Before I give you my answer I would like to
hear your answer to your own question; what do you think are the
important discoveries about mind and behavior that have been made
through documented empirical PCT research over the past 50 years?
Sorry: I asked first. _You_ tell _me_. (By the way, I did very much
appreciate Bill's reply reviewing the development of control theory and his
application of it to humans and other critters.)
Since yours is likely to be a very short answer to this question,
perhaps you might also add some examples of the important discoveries
about mind and behavior that have been made through documented
empirical _non-PCT_ research over the past 50 years.
Be careful what you ask for: you may just get it. To keep the list
manageable, I'll focus on only a subset of those that come to mind from some
of my own areas of interest.
Learning (acquisition) curve
Forgetting curve
Overlearning effect
Savings effect
Serial position effect
Primacy effect
Recency effect
Von Restorff effect
Proactive interference
Retroactive interference
Short-term memory
Chunking
Stroop effect
Dichotic listening results (selective attention)
Delayed auditory feedback effect
Serial scanning of lists in STM
Short-term habituation
Long-term habituation
Sensitization
Classical conditioning
Inhibition of delay
Disinhibition
Temopral conditioning
External inhibition
Extinction
Spontaneous recovery
Deprivation/satiation effects
Generalization
Discrimination
Peak shift phenomenon
Stimulus control
Concept formation
Reinforcement
Punishment
Avoidance
Conflict
Partial reinforcement extinction effect
Compound conditioning
Blocking
Overexpectation
Overshadowing
Second-order conditioning
Conditioned reinforcement
Chaining
Matching
Behavioral contrast
Spontaneous alternation
Transfer of training
Learned taste aversion
Conditioned suppression
Latent learning
Autoshaping
Priming
Observing behavior
Orienting response
Relevance of contextual stimuli
Attachment (Imprinting)
Critical (or sensitive) period
Separation anxiety
Vacuum activity
Fixed action patterns
Innate releasing stimuli
Flooding
Schedule effects
Potentiated startle
Neophobia
Instinctive drift
Learning set
Learned helplessness
Experimental neurosis
Self control
Hyperphagia
I could go on and on, but this is sufficient for illustration. Now, Rick,
how about your list?
Regards,
Bruce
[From Rick Marken (971223.0750)]
Bruce Abbott (971223.1020 EST) provides the following list of
"important discoveries about mind and behavior that have been
made through documented empirical _non-PCT_ research over the
past 50 years".
Learning (acquisition) curve
Forgetting curve
Overlearning effect
Savings effect
Serial position effect
Primacy effect
Recency effect
Von Restorff effect
Proactive interference
Retroactive interference
Short-term memory
Chunking
Stroop effect
Dichotic listening results (selective attention)
Delayed auditory feedback effect
Serial scanning of lists in STM
Short-term habituation
Long-term habituation
Sensitization
Classical conditioning
Inhibition of delay
Disinhibition
Temopral conditioning
External inhibition
Extinction
Spontaneous recovery
Deprivation/satiation effects
Generalization
Discrimination
Peak shift phenomenon
Stimulus control
Concept formation
Reinforcement
Punishment
Avoidance
Conflict
Partial reinforcement extinction effect
Compound conditioning
Blocking
Overexpectation
Overshadowing
Second-order conditioning
Conditioned reinforcement
Chaining
Matching
Behavioral contrast
Spontaneous alternation
Transfer of training
Learned taste aversion
Conditioned suppression
Latent learning
Autoshaping
Priming
Observing behavior
Orienting response
Relevance of contextual stimuli
Attachment (Imprinting)
Critical (or sensitive) period
Separation anxiety
Vacuum activity
Fixed action patterns
Innate releasing stimuli
Flooding
Schedule effects
Potentiated startle
Neophobia
Instinctive drift
Learning set
Learned helplessness
Experimental neurosis
Self control
Hyperphagia
Bruce continues:
I could go on and on, but this is sufficient for illustration.
Now, Rick, how about your list?
Ok. Here's my list:
Behavior is the control of perception
That's it.
You win on length of list. I win on number of actual discoveries
about mind and behavior. The score is: Rick - 1, Bruce - 0
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Bruce Abbott (971223.1335 EST)]
Rick Marken (971223.0750) --
You win on length of list. I win on number of actual discoveries
about mind and behavior. The score is: Rick - 1, Bruce - 0
I had a feeling I was going to loose, no matter what list I produced . . .
By the way, all those on my list are "discoveries" about mind and behavior.
How many of them would have been predicted by PCT (or HPCT)?
Here's my list:
Behavior is the control of perception
If you had said "control" I would agree that this has been empirically
established, for some variables at least. But the assertion that behavior
is the control of perception is theory (the core assumption of PCT, in
fact). That's why PCT is called perceptual control THEORY. Let's see if we
can get back on track. What empirical phenomena does PCT predict, which
might not be expected from other points of view?
Regards,
Bruce
[From Bill Powers (971223.1228 MST)]
Bruce Abbott (971223.1020 EST)--
Be careful what you ask for: you may just get it. To keep the list
manageable, I'll focus on only a subset of those that come to mind from some
of my own areas of interest.
A very impressive list. Before I comment on it, would you mind weeding it a
bit? What would the list look like if you just include the phenomena that
are seen in _every_ subject _every_ time the associated hypothesis is
experimentally demonstrated? (I'll settle for 95%).
Best,
Bill P.
[From Rick Marken (971223.1300)]
Bruce Abbott (971223.1335 EST) --
By the way, all those on my list are "discoveries" about mind
and behavior. How many of them would have been predicted by PCT
(or HPCT)?
Before I do that, I think we should determine which of your
"discoveries" are real phenomena.Some can surely be eliminated
by taking Bill Powers' (971223.1228 MST) advice and including only
those discoveries that are seen in 95% of the subjects "each time
the associated hypothesis is experimentally demonstrated". Others
can be eliminated because they are simply misinterpretations of
control phenomena. These include (but are probably not limited to):
Stimulus control
Reinforcement
Punishment
Avoidance
Conflict
Partial reinforcement extinction effect
Autoshaping
Orienting response
Fixed action patterns
Innate releasing stimuli
Schedule effects
Self control
I said that my list of discoveries about mind and behavior made
through documented empirical PCT research was as follows:
Behavior is the control of perception
You said:
the assertion that behavior is the control of perception is theory
Yes. And the discoveries about mind and behavior that have been made
through documented empirical PCT research over the last 50 years
show that the theory is correct: behavior _is_ the control of
various perceptual variables, including one and two dimensional
cursor position, size and shape of a quadrilateral figure, angular
velocity, sequence, relationship, etc. Demos like my "Test for the
Controlled Variable":
http://home.earthlink.net/~rmarken/ControlDemo/ThreeTrack.html
show that it is a (perceptual) representation that is controlled
(and not the physical variable itself). Try it and see if _you_
can figure out how we know that it is, in fact, a _perception_ of
the position of the squares, and not the position of the squares
itself, that is controlled.
What empirical phenomena does PCT predict, which might not be
expected from other points of view?
Why not start by looking at all the empirical phenomena at
http://home.earthlink.net/~rmarken/demos.html
All these phenomena are predicted by PCT (as seen by the behavior
of the PCT model in many cases) and by _no other_ "point of view".
By the way, thanks for that big long list of non-PCT "discoveries".
This was just a small subject of all the "discoveries" of non-
PCT psychology but it shows just how much you conventional
psychologists to protect. No wonder you guys don't like PCT.
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Bruce Abbott (971223.1720 EST)]
Bill Powers (971223.1228 MST) --
Bruce Abbott (971223.1020 EST)
Be careful what you ask for: you may just get it. To keep the list
manageable, I'll focus on only a subset of those that come to mind from some
of my own areas of interest.
A very impressive list. Before I comment on it, would you mind weeding it a
bit? What would the list look like if you just include the phenomena that
are seen in _every_ subject _every_ time the associated hypothesis is
experimentally demonstrated? (I'll settle for 95%).
I'd do this for you, if only I understood what you mean by "the associated
hypothesis is experimentally demonstrated." These are replicable empirical
phenomena, not hypotheses or tests of theories. 95% of what?
Regards,
Bruce
[From Rick Marken (971223.1445)]
Bill Powers (971223.1228 MST) --
A very impressive list. Before I comment on it, would you mind
weeding it a bit? What would the list look like if you just
include the phenomena that are seen in _every_ subject _every_
time the associated hypothesis is experimentally demonstrated?
(I'll settle for 95%).
Bruce Abbott (971223.1720 EST) --
I'd do this for you, if only I understood what you mean by
"the associated hypothesis is experimentally demonstrated."
These are replicable empirical phenomena, not hypotheses or
tests of theories. 95% of what?
95% of the subjects. So, for example, you have a "discovery"
called "learning curve". Do at least 95% of the individual
subjects in the experiment exhibit the "learning curve" that
was demonstrated?
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Bill Powers (971223.1621 MST)]
Bruce Abbott (971223.1720 EST)--
I'd do this for you, if only I understood what you mean by "the associated
hypothesis is experimentally demonstrated." These are replicable empirical
phenomena, not hypotheses or tests of theories. 95% of what?
I meant that the description of the phenomenon should fit what is observed
in every subject on every trial. For example, in serial memorizing, the
generalization might be that people remember more items from the beginning
and end of the list than from the middle. If 95% of the people show this
effect, or each individual shows it for 95% of the lists to be remembered,
this phenomenon can stay on the list of discoveries.
Or, if you like, you could just list the "statement of empirical
observation" followed by "true of x% of individuals in the study, or of
each individual on y% of the trials." You and I might have different ideas
about what constitutes a scientifically useful fact, so we can set our own
cutoff points.
Best,
Bill P.
[From Bruce Abbott (971224.1220 EST)]
Bill Powers (971223.1621 MST) --
Bruce Abbott (971223.1720 EST)
I'd do this for you, if only I understood what you mean by "the associated
hypothesis is experimentally demonstrated." These are replicable empirical
phenomena, not hypotheses or tests of theories. 95% of what?
I meant that the description of the phenomenon should fit what is observed
in every subject on every trial. For example, in serial memorizing, the
generalization might be that people remember more items from the beginning
and end of the list than from the middle. If 95% of the people show this
effect, or each individual shows it for 95% of the lists to be remembered,
this phenomenon can stay on the list of discoveries.
Or, if you like, you could just list the "statement of empirical
observation" followed by "true of x% of individuals in the study, or of
each individual on y% of the trials." You and I might have different ideas
about what constitutes a scientifically useful fact, so we can set our own
cutoff points.
I don't have that sort of information, but I can tell you that each of the
listed phenomena can be demonstrated in any normal individual of the
relevant species under the proper test conditions. Almost every one of them
was discovered in single-subject research. (The initial seven items on the
list are from Ebbinghaus, who tested himself.)
Your requirement is a strange one for a phenomenon. By it, one would have
to exclude such physical phenomena as thunderstorms (you don't get one on
95% or more of hot, muggy summer days) and blue eyes (which don't occur in
95% or more of people). People and animals don't always eat when they're
hungry, so eating would have to be excluded as too unreliable to be accepted
as a reliable phenomenon. Even for control, many CVs that are controlled
some of the time by some of the people are not controlled 95% or more of the
time or by 95% or more of people. Obviously not worth our attention, by
your criteria.
By the way, I presented my list only because Rick insisted that I do so.
Now, can be get back to talking about PCT?
Regards,
Bruce
[From Bill Powers (971224.1146 MST)]
Bruce Abbott (971224.1220 EST)--
I don't have that sort of information, but I can tell you that each of the
listed phenomena can be demonstrated in any normal individual of the
relevant species under the proper test conditions. Almost every one of them
was discovered in single-subject research. (The initial seven items on the
list are from Ebbinghaus, who tested himself.)
I can wait to see the actual numbers. Take your time.
By the way, I presented my list only because Rick insisted that I do so.
Funny, I thought the sequence began with your challenging PCTers to come up
with a list of discoveries. I asked my question because I wanted to know
what constitutes a discovery in your way of seeing things.
Now, can be get back to talking about PCT?
Sure.
Best,
Bill P.